일 | 월 | 화 | 수 | 목 | 금 | 토 |
---|---|---|---|---|---|---|
1 | ||||||
2 | 3 | 4 | 5 | 6 | 7 | 8 |
9 | 10 | 11 | 12 | 13 | 14 | 15 |
16 | 17 | 18 | 19 | 20 | 21 | 22 |
23 | 24 | 25 | 26 | 27 | 28 |
- NotFoundError
- 풀업저항
- TensorFlow
- Circuit
- pyTorch
- softmax backpropagation
- Softmax
- Big size image
- backward pass
- leetcode
- assignment
- error
- impl
- pulldown
- atmega128
- neural net
- 회로
- assignment1
- cs231n
- Features
- palindrome
- #9
- pullup
- autoencoder
- Solution
- 풀다운저항
- two-layer neural net
- Backpropagation
- KNN
- Floating
- Today
- Total
목록ML , DL (2019) (14)
코딩공부
![](http://i1.daumcdn.net/thumb/C150x150.fwebp.q85/?fname=https://blog.kakaocdn.net/dn/buqWf5/btqxGwOo4Eo/ZDQfk7suXrol4e3yK9BPO1/img.png)
지난글에 언급했듯이 Backpropagation, Intuitions 부분을 번역해보았다. Slide가 아닌 http://cs231n.github.io/optimization-2/ 에 있는 내용으로 Slide기준 Lecture 4에 포함된 내용이다. Slide의 일부도 필요에따라 끌어와 사용하였다. Introduction Motivation. In this section we will develop expertise with an intuitive understanding of backpropagation, which is a way of computing gradients of expressions through recursive application of chain rule.(of가 참 많다) Und..
![](http://i1.daumcdn.net/thumb/C150x150.fwebp.q85/?fname=https://blog.kakaocdn.net/dn/bDD1zt/btqxDVGdKJb/gqek76Usqn0K3CwzphYHH0/img.png)
softmax classifier는 svm과 내용이 상당히 비슷해서 SVM과 다른부분들만 설명을 하도록 하겠습니다. 설명이 없는 부분은 SVM과 같다고 생각하면 됩니다. 이번에도 코드는 github.com/Lee-daeho/cs231n 에 있습니다. 필요하신 분들은 가서 확인하시면 됩니다. scores를 구하는 부분은 SVM과 같다. 하지만 위의 TODO부분을 보면 If you are not careful here, it is easy to run into numeric instability. 라는 부분이 있다. 주의하지 않으면 numeric instability에 빠질 수 있다. 라는 의미인데 cs231n의 lecture note를 보면 다음과 같은 설명을 찾을 수 있다. Softmax function..
![](http://i1.daumcdn.net/thumb/C150x150.fwebp.q85/?fname=https://blog.kakaocdn.net/dn/wEPf6/btqxyKZQkw1/Anwan4V8lwNjN6VSPvViPK/img.png)
강의 내용 중 SVM에 관한 설명 부분을 번역해 보았습니다. There are several ways to define the details of the loss function. As a first example we will first develop a commonly used loss called the Multiclass Support Vector Machine (SVM) loss. The SVM loss is set up so that the SVM “wants” the correct class for each image to a have a score higher than the incorrect classes by some fixed margin Δ. Notice that it’s somet..
![](http://i1.daumcdn.net/thumb/C150x150.fwebp.q85/?fname=https://blog.kakaocdn.net/dn/dA8mXr/btqxtV1ozfT/eJTDS5mK0AdtRX2SW58Ke0/img.png)
글을 작성하기에 앞서 Ubuntu 환경에서 진행하였으며 anaconda와 같은 가상환경은 사용하지 않았다. cs231n의 강의를듣고 assignment1의 knn과제를 진행하였다. numpy에서 익숙치않은 함수들도 나와서 모두 이해하느라 어제부터 오늘까지 약 8시간은 걸린 것 같다. knn.ipynb파일과 k-nearest-neighbor.py파일은 https://github.com/Lee-daeho/cs231n/tree/master/assignment1 에 가면 확인할 수 있다. 블로그에서는 전체 내용이 아닌 일부분들을 번역하며 설명할 예정이니 전체 파일은 github에서 확인하기 바란다. cs231n의 assignment는 전체 코드의 일부분을 학생이 직접 코딩하여 전체 프로그램을 완성시키는 방식이다..